1. Introduction
In evidence-based policy (EBP), what is the role for mechanistic evidence—that is, information about the stepwise process in a causal chain by which a policy intervention causally produces an effect? Philosophers have focused on predicting effectiveness: If policymakers want to extrapolate reliably from the results “that it worked” in a randomized clinical trial (RCT) to novel domains, they need additional information about “how it worked.” An RCT shows only that “it worked there,” not that “it will work here.” To vouch for external validity, policymakers must identify the “support factors” for replicating an effective intervention (Cartwright and Hardie Reference Cartwright and Hardie2012). Without mechanistic evidence, their predictions are unreliable because they do not really know whether interventions will be robust and persistent across the population (Russo Reference Russo2012; Grüne-Yanoff Reference Grüne-Yanoff2016; Marchionni and Reijula Reference Marchionni and Reijula2019).
Existing philosophical accounts of EBP have sidelined or excluded politics, values, and context from their analysis. For instance, Cartwright and Hardie focus exclusively on empirical evidence and intervention effectiveness, in isolation from social factors like political expediency and choice of goals (Reference Cartwright and Hardie2012, 12). Yet we artificially insulate scientific epistemology at our own peril, as such idealization ignores the “epistemic friction” from social realities operative in policymaking, such as how commercial forces contribute to medical disasters by shaping the questions asked about mechanisms and the methods chosen for studying them (Holman Reference Holman2019). In public policy, resistance to paternalistic interventions like “nudges” often involves ethical concerns such as respect for autonomy (Rouyard et al. Reference Rouyard, Engelen, Papanikitas and Nakamura2022; Grüne-Yanoff Reference Grüne-Yanoff2016).
I argue that mechanistic evidence itself can be a political instrument of social power, which science advisers use to support or undermine contested treatments (ChoGlueck Reference ChoGlueck2019, Reference ChoGlueck2021). Rather than limiting our analysis of mechanisms to merely predicting effectiveness, it is important to broaden the scope of our philosophy to cover the moral, ethical, religious, and political dimensions of mechanistic claims. When it comes to the actual practice of policymaking, these political dynamics cannot simply be ignored. For instance, during the COVID-19 pandemic, the effectiveness of coronavirus interventions depended on family structures, overcrowding, beliefs about vaccine safety, and distrust of medicine (Aronson et al. Reference Aronson, Daniel Auker-Howlett, Kelly and Williamson2021, 690). These “social and psychological factors are key to uptake of and adherence to an intervention,” and they “can change radically during an epidemic” (Aronson et al. Reference Aronson, Daniel Auker-Howlett, Kelly and Williamson2021, 691; Goldenberg Reference Goldenberg2021).
I develop a more nuanced account of how the “evidence” in policymaking can be both empirical and political, contrary to the rational–social dichotomy (Longino Reference Longino2002). I compare two strategies: The wedge is a stalling technique, where opponents of a policy change identify a potentially fatal flaw in the proposed causal chain and call for more empirical research. The buttress preempts such stalling, where advocates foresee empirical weaknesses and conduct premediated research. Focusing on reproductive health policy, I analyze how those committed to “pro-life” and “pro-choice” agendas have used evidence of mechanisms as political instruments for/against access to contested interventions of contraception and abortion pills. Ethically speaking, the pro-life/pro-choice binary is also a false dichotomy that fails to represent the complex social issues surrounding reproductive liberty, particularly for poor women and women of color (Ross Reference Ross2006). Nonetheless, the back-and-forth dynamic between these dueling camps allows us to capture the interplay between the science and politics of reproductive health policy.
This article contributes to philosophy of science policy about behavioral mechanisms and the politics of doubt. First, regarding mechanism, I explain how debates over what Cartwright and Hardie call support factors are mired in politics, values, and background assumptions. Although philosophers have recognized mechanistic knowledge’s pragmatic value, we have mostly focused on biochemical mediators involved in treatment mechanisms (Craver and Darden Reference Craver and Darden2013; ChoGlueck Reference ChoGlueck2019) rather than behavioral and cognitive factors (Grüne-Yanoff Reference Grüne-Yanoff2016). Delving into the larger set of support factors for a drug mechanism, we see debates among policymakers over how patient behaviors like compliance/noncompliance with instructions might mediate the causal pathway. Mediators are the intermediate steps between a treatment and its outcomes. Policymakers also dispute moderators, support factors that amplify or diminish a treatment’s effect, such as how gestational age could be an independent variable that moderates drug effectiveness (for more on causal diagramming, see Marchionni and Reijula Reference Marchionni and Reijula2019). Because these sociobehavioral dimensions of drug mechanisms are understudied, they are ripe for political contestation.
Second, regarding the politics of doubt, this article also contributes to philosophy of science policy by contrasting different approaches of advocacy involving the manipulation of uncertainty via information related to mechanisms. Philosophers and historians have elucidated the techniques of “the tobacco strategy,” which other industries later used to undermine regulations by exploiting scientific uncertainties with a “politics of doubt” (Fernández Pinto Reference Fernández Pinto2017; Oreskes and Conway Reference Oreskes and Conway2010). Likewise, this article presents two case studies in which advocates “play up” empirical uncertainty or “nail down” robust effectiveness: Section 2 analyzes how mechanistic evidence can function as a wedge. Here, pro-life critics of emergency contraception used mechanistic plausibility to cleave US Food and Drug Administration (FDA) policies apart from empirical support. Section 3 then shows how mechanistic evidence can function preemptively as a buttress. There, pro-choice advocates of medication abortion used mechanistic evidence to shore up support with new empirical studies. Section 4 explains why neither strategy is foolproof for achieving political aims.
2. The wedge strategy
The wedge involves three features: (1) attempted deadlock, (2) targeting a weak link, and (3) conjecturing implausibility. First, the wedge is a means of obstructing one’s political opponents to stymie their proposed policy changes, thus favoring the status quo. Second, the wedge targets a potentially weak part of the proposed/assumed causal chain that is required for the policy to work effectively—framed as a purely scientific concern about the lack of evidence for a mechanism but based in part on ethical/political value judgments. Third, while the wedge is contested by the opposition as politically motivated speculative fiction, it appeals to background assumptions that could be evidenced but do not presently have direct empirical support. Thus, the wedge employs a debatable standard of evidence based on mere biological plausibility rather than tested actuality. Although some philosophers, such as Hansson (Reference Hansson2014), have assumed that mere plausibility is all that mechanistic investigations can provide, we will see how a stronger role of establishing mechanistic robustness is also possible (Grüne-Yanoff Reference Grüne-Yanoff2016; Cartwright and Hardie Reference Cartwright and Hardie2012).
The first case involves socially conservative pro-life science advisers, who posited that the lesser cognitive abilities of adolescent girls would undermine their compliance with label instructions (Stanford et al. Reference Stanford, David Hager and Crockett2004). Although much pro-life advocacy is aimed directly against abortion, pro-life advocates have also opposed hormonal forms of birth control (Williams Reference Williams2016; Miller Reference Miller2015). My previous work shows how they falsely equate emergency contraceptive pills with abortion, based on alleged postfertilization mechanisms (ChoGlueck Reference ChoGlueck2019, Reference ChoGlueck2022a). Often framed as religious opposition to hormonal contraception, pro-life advocates aimed to drive a wedge in the scientific evidence that supported increased contraceptive access, and they successfully delayed the switch to over-the-counter (OTC) emergency contraceptives for nearly a decade (Prescott Reference Prescott2011).
In 1999, the FDA approved the emergency contraceptive pill Plan B for prescription sale. This progestogen-based medication involves 1.5 mg of the hormone levonorgestrel (LNG), which is 84% effective if taken within 72 hours of unprotected sex. LNG prevents pregnancy by preventing the release of an egg from the ovaries (Endler et al. Reference Endler, Li and Gemzell Danielsson2022). The biochemical mechanism of action involves three mediator steps:
-
1. Orally ingesting 1.5 mg of LNG delays or inhibits the luteinizing hormone peak, but only if taken prior to ovulation—during the follicular phase when the lead follicle in the ovaries is still developing (Marions et al. Reference Marions, Hultenby, Lindell, Sun, Ståbi and Gemzell Danielsson2002, Reference Marions, Cekan, Bygdeman and Gemzell-Danielsson2004; see also Gemzell-Danielsson Reference Gemzell-Danielsson2010, fig. 1).
-
2. This chemical change delays the subsequent luteal phase by stopping the lead follicle from fully maturing into the corpus luteum, which would have produced the progesterone and estradiol necessary for pregnancy (Croxatto et al. Reference Croxatto, Brache, Pavez, Cochonb, Forcelledoa, Alvarez, Massaia, Faundesb and Salvatierra2004; Brache et al. Reference Brache, Cochon, Deniaud and Croxatto2013).
-
3. This delay in follicular rupture leaves the egg immature and unreleased from the ovary, disabling fertilization potential and thus causing a contraceptive effect (Ortiz et al. Reference Ortiz, Ortiz, Fuentes, Parraguez and Croxatto2004; Noé et al. Reference Noé, Croxatto, Salvatierra, Reyesa, Villarroela, Muñoza, Moralesa and Retamales2011).
Drug mechanisms are often conceptualized in biochemical terms like these. Yet once a prescription-only drug becomes available OTC and medical oversight is relaxed, other dimensions of its mechanism of action can become more salient, particularly the cognitive requirements for patient understanding and compliance.
After FDA approval, the feminist-run drug company Women’s Capital Corporation applied for an OTC switch. This ushered in a major controversy for the FDA under the conservative pro-life administration of George W. Bush, who opposed contraception along with abortion on the assumption that the former results in the latter (Prescott Reference Prescott2011). During the public hearings in 2003, opponents of Plan B claimed that young women were undisciplined, unreliable users because of their “inability to control themselves in sexual situations … these women are slaves to their bodies” (FDA 2003, 239; Wynn and Trussell Reference Wynn and Trussell2006). This thinly veiled sexism assumed that the users of Plan B were untrustworthy and unlikely to conform to conservative norms for “good women” to accept their maternal duty and bear fertilized zygotes to term (ChoGlueck Reference ChoGlueck2022b).
In the deliberation of OTC approval, 5 of the 27 committee members voted against the switch, claiming that younger age groups were less responsible and would not comply without the medical oversight of a prescriber. One of these dissenters was David Hagar, a Bush appointee infamous for refusing to prescribe contraceptives to unmarried women on the basis of religious grounds and “family values” (McGarvey Reference McGarvey2005). Hagar contested the generalizability of the RCT to 9- and 10-year-old girls and suggested an age restriction (FDA 2003, 370; Wynn and Trussell Reference Wynn and Trussell2006, 303). Other experts balked at this conjecture because menarche, the age of one’s first menstrual cycle, is closer to 12 years old (FDA 2003, 371).
Both supporters and detractors of Plan B questioned the cognitive abilities of young women, yet only the latter took such to undermine the legitimacy of an OTC switch. For instance, one pediatrician claimed that because of “inherent aspects of adolescent development,” young girls were “not good at planning ahead” and “poor contraceptors … [who thus] need access to emergency contraception” (FDA 2003, 388). This belief that adolescents might be less mature developmentally was echoed in the inverse judgment by Susan Crockett, who served on the board of the American Association of Pro-Life Obstetricians and Gynecologists (Wynn and Trussell Reference Wynn and Trussell2006, 303). She worried about extrapolation because of the cognitive differences and voted against Plan B’s application, urging “testing of pediatric use” (FDA 2003, 378). Crockett and Hagar convinced the committee chair that “the population that we have extremely little information on [is] the adolescents,” and thus he recommended “a study could be done to assess their behaviors” to determine whether or not to impose an age restriction (FDA 2003, 386).
Looking back to our three-step mechanism and the three features of the wedge, we see how opponents of Plan B contested the OTC policy switch: The pro-life camp made a plausible but politically motivated strike against the extrapolation that younger adolescent girls (aged 9–13) would likely have lower compliance as a result of their supposed imprudence. They posited that age was a behavioral moderator of the first step in the mechanism—where ingesting LNG inhibits the luteinizing hormone peak but only if ingested prior to ovulation—such that it could weaken patient compliance without “mature” cognitive abilities.
The following spring, the acting director of the FDA’s Center for Drug Evaluation and Research (CDER), Steven Galson, denied OTC approval. He claimed there were “large developmental differences” between older and younger adolescents and inadequate data for extrapolation; only 5% of the subjects were 14–16 years old, and no participant was <14 (Galson Reference Galson2004). Galson’s position echoes philosophers’ caution against extrapolation via simple inductive inference when interventions operate through different mechanisms in the base population compared with the target population (Steel Reference Steel2008). Yet a follow-up investigation by the congressional watchdog characterized Galson’s reasoning as “novel” and “atypical” because no other FDA-approved contraceptive has had an age restriction or required pediatric studies (US Government Accountability Office [GAO] 2005, 31, 32). Moreover, the trial had found that among subjects aged ≤16 (n = 29), there were actually lower rates of incorrect use, pregnancy, and adverse events (Raymond et al. Reference Raymond, Chen and Dalebout2003, 18, table 3).
The FDA eventually approved OTC access in 2006, but with a restriction that required consumers to show proof of age ≥18. Plan B’s restriction was only removed by a 2013 court order, in which the judge considered the evidence from adolescents relevant but unnecessary: “If the actual use study is essential, it is only because the FDA failed to follow its practice of extrapolating data from older adolescents and also departed from its policy of relying principally on the product label to result in appropriate use” (Tummino v. Hamburg, 2013, 5; Rovner Reference Rovner2013). Although it is reasonable to posit that cognitive abilities are a moderator of patient compliance, it is questionable whether this actually undermines extrapolation below 16 and warrants restricting OTC access for everyone <18. Nonetheless, socially conservative antiabortion advisers found these value-laden assumptions plausible and actionable.
Now, one might question: Is this wedge really politically motivated? It could be that these advisers and directors are merely acting on the state of the evidence and thus making value-free judgments about extrapolation. I find this interpretation unpersuasive, given the unprecedented, top-down nature of FDA decisions during the Bush administration, with its opposition to reproductive health initiatives (Prescott Reference Prescott2011; Carpenter Reference Carpenter2010; Waggoner Reference Waggoner2017). It is closer to a “politics of doubt,” in which scientific uncertainty is weaponized to oppose contraceptive access through unachievably high standards of evidence (Davidoff and Trussell Reference Davidoff and Trussell2006). For instance, Susan Wood, then the director of the FDA Office of Women’s Health, advocated “strongly against introducing subtle barriers to access, such as raising flags about age” (FDA 2003, 393). She later resigned from her post because the agency was “disregarding the scientific and clinical evidence,” compromising access to a drug that would “reduce the rate of unplanned pregnancies and the need for abortions” (Wood Reference Wood2005, 1650).
Although Wood is correct about the political nature of the wedge, this does not necessarily entail its being un-empirical, unless one accepts the rational–social dichotomy (Longino Reference Longino2002): The wedge strategy works by appealing to background assumptions that are testable in principle but are simply taken for granted. The advocates of Plan B did, in fact, later test and disconfirm their assumption about age moderating the mechanism by evaluating whether girls aged 12–17 understood key concepts from the drug label, for example, that LNG should be taken as soon as possible after unprotected intercourse (Raymond et al. Reference Raymond, L’Engle, Tolley, Ricciotti, Arnold and Park2009). Although younger adolescents aged 12–14 did have significantly less understanding than those aged 15–17, the researchers found that more than four out of five of these younger girls understood all of the major concepts. These researchers concluded that all “female adolescents understand the prototype package label well enough to enable safe and effective use without assistance from a clinician” (Raymond et al. Reference Raymond, L’Engle, Tolley, Ricciotti, Arnold and Park2009, 199). Thus, this wedge is empirical in the sense of being a testable conjecture (and later disconfirmed). Yet the goal of the wedge is not truth in itself but rather preventing policy change by contesting the evidence at hand.
3. The buttress strategy
The second mechanistic strategy, the buttress, involves three features that subvert a wedge: (1) anticipating deadlock, (2) testing a weak link, and (3) attempting to refute implausibility. First, whereas the wedge is reactive, the buttress is a preemptive means of preventing obstruction by one’s political opponents to enable policy change, thus disfavoring the status quo. Second, the buttress targets a potentially weak part of the proposed causal chain required for a policy’s effectiveness, but it does so to fortify rather than shake it. Third, the buttress strategy provides more direct evidence than the wedge for the actual robustness of the support factors (moderator, mediator, etc.) under question. The role of ethical/political values here is not setting the standard of evidence per se but rather motivation to pose the question and collect the data.
In our second case study, mechanistic evidence was used as a prophylactic tool by socially liberal abortion advocates, including Planned Parenthood and the American College of Obstetrics and Gynecology (ACOG). To stave off antiabortion pro-life attacks, pro-choice advocates conducted research to generate mechanistic evidence of direct-to-patient delivery of the abortion pill, which then accounted for two out of three abortions in the United States (Jones and Friedrich-Karnik Reference Jones and Friedrich-Karnik2024). As we shall see, the TelAbortion Project demonstrated the empirical equivalence of two different means of prescription delivery (in person vs. teleconference) for ensuring the proper timing and dosing needed to result in the subsequent biochemical steps, buttressing behavioral support factors (mediators of patient compliance).
In 2000, the FDA approved mifepristone (RU-486) for a nonsurgical abortion in the first 10 weeks of pregnancy. The updated standard treatment for medication abortion up to week 12 is 200 mg of mifepristone followed by 800 μg of misoprostol, which is FDA approved for treating stomach ulcers and commonly prescribed off-label for inducing labor (World Health Organization [WHO] 2022). Although medication abortion is possible up to week 14, the WHO recommends a surgical abortion of dilation and evacuation in the second trimester.
Mifepristone and misoprostol induce a miscarriage by initiating contractions and expelling the embryo from the uterus. The biochemical mechanism of action involves the following four steps:
-
1. First, a patient in their first trimester ingests 1 tablet of mifepristone, which is an antiprogestin that binds to progesterone receptors in the uterus (Hagey et al. Reference Hagey, Givens and Bryant2022).
-
2. This molecular change stops the pregnancy from growing by outcompeting endogenous progesterone and blocking its action in the endometrial cells, which results in vascular changes like bleeding, injury, and cell death (Johannisson et al. Reference Johannisson, Oberholzer, Swahn and Bygdeman1989).
-
3. Then, 24 to 48 hours after mifepristone, the patient takes 4 tablets of misoprostol, orally or vaginally, which degrades collagen in the cervix, softening and thus dilating it (Bakker et al. Reference Bakker, Pierce and Myers2017).
-
4. Mifepristone sensitizes the smooth muscles in the myometrium and results in uterine contractions, such that misoprostol can induce labor quickly and empty the uterus of pregnancy contents (Gemzell-Danielsson et al. Reference Gemzell-Danielsson, Bygdeman and Aronsson2006).
Note the implicit importance of social and behavioral support factors: gestational-age eligibility and patient compliance with instructed timing.
When the FDA approved mifepristone in 2000, it determined that there were risks of serious complications, so it created a Risk Evaluation and Mitigation Strategy (REMS) program for safety, although scholars have contested this special paternalism as unnecessary and unwarranted (Paradise Reference Paradise2023). Most importantly, the REMS required in-person dispensation directly to patients; this prohibited the mailing of mifepristone (FDA CDER 2025).
During the early 2010s, the pro-life movement’s attacks through state laws resulted in a significant decline in the national rate of abortions (–14%) and the number of clinics providing abortions (–6%) (Jones and Jerman Reference Jones and Jerman2017). Accordingly, to assess and expand virtual provisioning of medication abortion, pro-choice researchers at Planned Parenthood and its research arm, the Guttmacher Institute, began the TelAbortion Project (Raymond et al. Reference Raymond, Chong, Winikoff, Plataisa, Marya, Lotarevicha, Castillo, Kaneshiro, Tschann, Fontanilla, Baldwin, Schnyer, Coplon, Mathieu, Bednareke, Keady and Priegue2019). This project compared direct-to-patient delivery with in-person distribution, using videoconference to determine eligibility based on results from lab tests, including ultrasounds. If eligible, patients were mailed medications with instructions and scheduled for a follow-up pregnancy test. The researchers found comparable success and safety rates, with around 94% of abortions completed and 1% reporting serious adverse effects. This buttressing evidence was replicated by another study conducted by Planned Parenthood and the ACOG, published the same year (Kohn et al. Reference Kohn, Snow, Simons, Seymour, Thompson and Grossman2019).
Looking back to the four-step mechanism and the three features of the buttress, we can see how proponents of medication abortion shored up mechanistic evidence for this alternative mode of delivery. Crucially, their buttress aimed to fortify the behavioral and social conditions for properly timed ingestion: Gestational age is a known moderator for effectiveness, with ∼10 weeks being the traditional cutoff; providers used to require that patients come to the office in person to determine their eligibility for medication abortion (as opposed to surgical approaches in later trimesters). Pro-choice proponents focused on steps 1 and 3—which require temporal coordination with gestation—to show that direct-to-patient delivery would not compromise effectiveness or safety. The TelAbortion Project assessed gestational age over videoconference, and thus these researchers buttressed the behavioral mediators related to patient compliance to show direct-to-patient delivery was equally safe and effective as in-person delivery of the prescription. And their motivation was straightforwardly political in terms of social liberalism and egalitarianism: to combat pro-life opposition and increase access through a wider means of provisioning.
The episode that followed during the COVID-19 pandemic illustrates the dynamics between the two strategies, in which antiabortionists tried to use a wedge, only to find they had been preempted by a buttress. When the pandemic severely limited access to face-to-face care, the ACOG sued the FDA to drop the in-person requirement, which the agency did in December of 2021 “to reduce burden on the health care delivery system” (FDA CDER 2025). Pandemic research further buttressed the direct-to-patient delivery, showing ultrasounds were unnecessary despite gestational age being a moderator (Anger et al. Reference Anger, Raymond, Grant, Haskell, Boraas, Tocce, Banks, Coplon, Shocheta, Plataisa and Winikoff2021). The following year, several pro-life organizations sued the FDA for its 2000 approval of mifepristone and its pandemic rule changes to the REMS. Surprisingly, although the Supreme Court had just overturned Roe v. Wade, it ruled unanimously that antiabortion organizations did not have standing to sue because their members refused to prescribe abortion pills and thus could not be injured by the FDA (VanSickle Reference VanSickle2024).
This episode illustrates the political effectiveness of the buttress as well as its limitations: Pro-choice researchers were able to persuade the FDA to drop their requirement using empirical evidence of behavioral mechanisms. Although the agency has traditionally seen mechanistic evidence as optional and secondary to RCTs, the FDA does take mechanisms seriously at times—such as when a mechanism is novel and the drug company appeals to mechanistic innovation in its application (Carpenter Reference Carpenter2010, 397–404; Holman Reference Holman2019). However, unlike the FDA, the Supreme Court’s decision was less determined by mechanistic evidence and more by legal considerations related to standing and injury. This suggests that the “evidence” component of EBP is still limited in US health policy, particularly for the judiciary.
Nonetheless, it would be misleading to cast the buttress strategy as purely empirical—and the courts as empirically negligent—because it is politically motivated evidence: Antiabortion scrutiny could potentially fall on the mode of delivery, specifically the potential for improper timing/dosing without direct medical oversight. So, as a bulwark against pro-life restrictions, pro-choice researchers delved deeper into the potentially weaker parts of the purported causal chain. This shows the anticipatory motivations to evaluate the behavioral mechanism, given that the desire to buttress the mechanistic evidence would increase access to abortion and bolster reproductive autonomy. This is rooted in the ethical valuation of the liberal right to abortion and the egalitarian importance of equal access for all, particularly during the pandemic, when many fell through the cracks of an overtaxed medical system.
4. Discussion
These strategies show how the empirical and political dimensions of mechanistic evidence interact through social factors (like standards of evidence) and cognitive factors (like motivated reasoning) (Longino Reference Longino2002). Now, one might ask: In what sense are each of these mechanistic strategies empirical? I contend that the buttress strategy is empirical in a stronger sense than the wedge because the latter is testable in principle, whereas the former has actually been tested. Recall that the wedge against Plan B attempted to undermine the evidence of effectiveness by positing the inability of young adolescent girls to control their sexuality and follow the instructions on the box without medical oversight. The obstruction was based on a resistance to extrapolate downward from data on young adolescents aged 14 and older, and the assumption was eventually disconfirmed via a comprehension study that included 164 girls aged 12–14 (Raymond et al. Reference Raymond, L’Engle, Tolley, Ricciotti, Arnold and Park2009). In contrast, the pro-choice buttress aimed to more positively shore up evidential support so that the pro-choice side’s arguments would be more plausible empirically when they were eventually contested by something like a pro-life wedge. Thus, because the buttress’s empirical nature is based on confirmed actuality rather than speculated possibility, it has a stronger sense of being empirical evidence than the wedge. This casts doubt on Hansson’s claim (Reference Hansson2014) that mechanistic evidence is only about conjecturing plausibility rather than also positively evidencing mechanistic robustness and persistence (Grüne-Yanoff Reference Grüne-Yanoff2016; Cartwright and Hardie Reference Cartwright and Hardie2012).
Another potential concern is whether there is anything specifically mechanistic about the wedge and buttress. It could be that these strategies are less related to mechanisms per se and more about politicizing scientific uncertainty, given that mechanisms are treated as “less certain” evidence. In response, I do see a sense in which the wedge is continuous with the politics of doubt used in other cases to magnify scientific uncertainty for political ends, such as with the tobacco strategy (Fernández Pinto Reference Fernández Pinto2017). However, I maintain that both strategies are chiefly mechanistic because each contests how a policy intervention will or will not work by targeting a weakness in support factors (moderator/mediator in the causal chain), either to undermine plausibility (wedge) or for direct testing (buttress).
Philosophers might also wonder how generalizable these strategies are beyond the pro-life/pro-choice dynamic. Although my evidence is limited, I do not think they are unique to this context, particularly the wedge: In the back-and-forth over COVID vaccines in 2021, the antivaccination camp used a wedge against these mRNA vaccines based on how they might plausibly work—namely, spreading fear that this novel form of vaccination would change or alter the genome or otherwise “interfere with human DNA” (Hammarlin et al. Reference Hammarlin, Kokkinakis, Miegel, Stoencheva, Borin, Hammarlin, Kokkinakis and Miegel2024, 160). The buttress is likely less common because it involves both the anticipation of deadlock and the empirical refutation of implausibility, which require intimate knowledge of one’s opposition and access to scientific expertise. Although this sophistication is a feature of the pro-choice movement, other camps counter wedges too late after the fact. Such was the case with the medical establishment’s “counter-epistemic take on mRNA rumors,” including the backfire from a 2022 study that concluded “we do not yet know if the converted DNA [from the vaccine’s mRNA] is integrated into the cells’ DNA,” inadvertently legitimizing the plausibility of the feared gene-therapy mechanism (Hammarlin et al. Reference Hammarlin, Kokkinakis, Miegel, Stoencheva, Borin, Hammarlin, Kokkinakis and Miegel2024, 161). Furthermore, there is nothing inherently “conservative” about the wedge or “liberal” about the buttress. For example, liberals led early pushback to artificial intelligence (AI) in schools with a wedge, contesting the plausibility of purported learning benefits due to skill loss (Elsen-Rooney Reference Elsen-Rooney2023).
5. Conclusion
Mechanistic evidence can function like a political instrument. For the wedge, policy-change opponents seek to break the proposed causal chain by conjecturing alternative conditions, such as adolescent noncompliance, under which moderators will diminish the mechanism’s effect. And its ultimate goal is to preserve the status quo, to stymie policy change through doubt. For the buttress, policy-change supporters reinforce the proposed chain epistemically by demonstrating empirically that even under alternative conditions, such as mail-order delivery, we can still count on safety and effectiveness. Thus, the buttress is a strategy of change from the status quo. Understanding the dynamics between these two strategies can help us better evaluate their empirical and political dimensions, both to ground our policies in scientific evidence and to critically evaluate the assumptions, motivations, and values behind alleged evidence of mechanisms.
Acknowledgments
This article was occasioned by our Philosophy of Science Association (PSA) symposium’s focus “Behavioral Mechanisms in Public Policy: What They Are, Why They Matter, and How to Detect Them” (held in New Orleans on November 15, 2024). I thank Till Grüne-Yanoff for inviting me and my fellow contributors for their generous feedback on my talk and manuscript: Caterina Marchionni, Yavor Paunov, Claudia Schwirplies, and Daphne Truijens. I also received helpful feedback from Edouard Machery and my fellow fellows at the University of Pittsburgh’s Center for Philosophy of Science: Margret Farrell, Mark Risjord, Marta Bielinska, Márton Gomori, Laurenz Casser, David Thorstad, and Kareem Khalifa.
Funding statement
None to declare.
Declarations
None to declare.
